Radioiodinated monoclonal antibodies are important for the diagnosis and therapy of cancer as summarized by Goldenberg in Amer. J. Med. 94:297-312 (1993). A number of methods have been developed over the last thirty years to chemically introduce radioiodine into monoclonal and polyclonal antibodies for these uses. Iodine is preferred as a radiolabel in these applications because the chemistry used for radioiodination of protein is relatively easy, radioiodine has useful physical decay characteristics, and isotopes of iodine are commercially available.
Among useful iodine isotopes are Iodine-124, which has been used to radiolabel antibodies as described by Pentlow et al., Journal of Nuclear Medicine, 37:1557-62 (1996), and Iodine-125, which has been used for detection using an intraoperative probe as described by Martin et al., Cancer Investigation, 14:560-71 (1996). In the context of using these iodine isotopes, one concern is the long circulation time of radioiodinated antibodies, which leads to high background radiation. The high background problem is compounded by the loss of radioiodine from target cells, when standard radioiodination methods are used. A poor target to non-target ratio of delivered iodine often results from the high background and radioiodine loss problems. Accordingly, a principal aim in the art is to improve the target to non-target ratio. Iodine-125 has been proposed for therapy purposes because of its cascade Auger electrons as described by Aronsson et al., Nuclear Medicine and biology, 20:133-44 (1993). Clearly, optimum use of a long-lived [t1/2, 60 days] low energy-emitting nuclide demands that intracellular target retention be achieved; which is not possible with conventional radioiodination methods.
Various chemistries have been developed to link iodine to antibodies that target cancer cells. These chemistries have been reviewed by Wilbur, Bioconjugate Chemistry 3:433-70 (1992). The most common linking procedure has been to prepare in situ an electrophilic radioiodine species to react with a functional group on an antibody. Reagents such as chloramine T and iodogen have been employed to generate electrophilic iodine. A tyrosine group on protein is usually the site of iodination. However, the presence of a harsh oxidant or reductant may lead to structural impairment of an antibody. For this reason, an alternative approach is to iodinate a small organic molecule and couple the pure iodinated species to antibody. N-Succinimidyl 3-(3-iodo-4-hydroxyphenyl) propionate (Bolton-Hunter reagent) is an example of the latter category. These and other methods have been reviewed by Wilbur (Id).
A major drawback with using the foregoing radioiodination schemes is the phenomenon of in vivo deiodination. As a result of antibody internalization and lysosomal processing in vivo, a labelled protein is degraded to small peptides, and its radioiodine is released from the cell in the form of iodotyrosine or as iodine attached to a low molecular weight peptide fragment. These findings have been reported by Geissler et al., Cancer Research 52:2907-2915 (1992) and Axworthy et al., J. Nucl. Med. 30:793 (1989). Such in vivo removal of radioiodine from target cells has a profound bearing on the use of iodine isotopes for radiodiagnosis and radiotherapy. Discrimination between tumor and non-tumor that is relevant to diagnosis and therapy, and the prolonged retention of isotope on a tumor cell, relevant to radiotherapy, are severely compromised by the occurrence of in vivo deiodination. This is readily appreciated if one considers the 8-day half-life of iodine-131, which is widely used for radioimmunotherapy investigations. If antibodies radioiodinated with this isotope are metabolized with consequent removal of the isotope from the target cells within the first 24-120 h post-injection of the reagent, the advantage of the lengthy half-life of this isotope for therapy is lost. That is, the useful half-life of this isotope is not exploited in a prolonged tumoricidal effect because of the above-described drawback of conventional radioiodination chemistry.
In contrast to this drawback of conventional chemistry, the action of in vivo deiodinases in releasing iodine in the form of molecular iodine from the cell is less significant to the problem of optimizing the target to non-target ratio of radioisotope accumulation. Workers in this field have attempted to prepare iodinated proteins that do not ‘deiodinate’ by the action of in vivo deiodinases as reviewed by Wilbur et al., Journal of Nuclear Medicine, 30:216-26 (1989). But these attempts have failed to show improvement in cellular retention of radioiodine. The reason for these failures is that, contrary to what was expected, the metabolic clearance of intact iodotyrosine was more important to clearance of isotope than was the deiodination of tyrosine to liberate radioiodine.
One way to overcome the unacceptably fast release of radioiodine from conjugate is to attach iodine to non-metabolizable carbohydrate and to conjugate the resultant entity to antibodies. After antibody catabolism within a tumor cell, the radioiodine remains stably attached to the carbohydrate and thus is trapped inside the cell. These carbohydrate labels, referred to as ‘residualizing labels’, are exemplified by Strobel et al., Arch. Biochem. Biophys 240:635-45 (1985) and Ali et al., Cancer Research (suppl) 50:783s-88s (1990). However, these methods, when applied to the labelling of monoclonal antibodies (Mabs), suffer from one or both of the following drawbacks: (1) Very low radiolabeling yields (3-6%) and (2) formation of aggregates (up to 20%). Low conjugation yield necessitates handling a large amount of radioactive iodine to incorporate sufficient radioactive label in antibody. This approach causes a radiation safety concern as well as wastage of most of the unusable radioactivity. As a result, the specific activity achieved by this method suffers. Furthermore, aggregate formation can lead to reduced tumor uptake and will lead to enhanced liver uptake, and thereby impair the effectiveness of the radiolabel method. The full advantage of using residualizing labels for radioimmunodetection and radioimmunotherapy cannot be realized unless progress can be made to limit the twin problems of poor radiolabeling yield and aggregate formation when using carbohydrate-based reagents. Using novel substrates and methodologies to address these issues is another aspect of this invention.
One well known approach to this problem is to label the antibody with a radiometal ion such as indium-111 or an isotope of yttrium, using a bifunctional aminopolycarboxylate ligand such as bifunctional EDTA or bifunctional DTPA. These radiolabelled conjugates exhibit prolonged retention of radiometal in tumor as exemplified in in vivo animal experiments by Stein et al., Cancer Research 55:3132-39 (1995). That is, radiometal ions chelated to aminopolycarboxylates also behave, in vivo, as residualizing labels. Thus, the problem of residualization generally applies to techniques that use these labels as well.
The prior art has addressed the issue of residualizing iodine labels by using non-metabolizable sugars to which an iodinatable group is attached. An iodinatable group such as tyramine is reductively coupled to the carbohydrate, so that there is no metabolizable peptide bond between tyramine and the sugar entity. There are two main problems encountered with these prior art methods. These are in the antibody-coupling steps. One method, that of Strobel et al. (see above), uses a carbohydrate-adduct derived from lactose, and couples proteins and antibodies to the same by first oxidizing the galactose portion of such adducts with galactose oxidase. Usually poor overall yield (3-6%) is obtained, as described by Stein et al. Cancer Research, 55:3132-3139, (1995). Furthermore, lactose is an inefficient substrate for galactose oxidase. In examining a number of galactose-containing carbohydrate derivatives for their ability to be oxidized by this enzyme, Avigad et al. (J. Biol. Chem 237:2736-2743, (1962)), determined that lactose had less than half the affinity of D-galactose for galactose oxidase, and was oxidized fifty times slower compared to galactose. This inefficient step therefore contributes to overall reduced radioisotope incorporation into antibodies.
Another approach involved in coupling to antibodies does not make use of any special property such as the ability of the carbohydrate to be selectively derivatized by an enzyme (such as galactose oxidase oxidation involving galactose moiety), but makes use of cyanuric chloride as the cross-linker to link both the iodinated carbohydrate and antibody. This approach has the serious problem of generating antibody aggregates. Cyanuric chloride has been used to form conjugates but unfortunately this reagent contains three reactive chlorines and consequently forms aggregates. Another factor involved in aggregate formation is the presence of multiple amino residues in antibodies that can bind to the residualizing agent and/or coupling reagent, particularly with carbohydrate residualizing agents that couple to protein by reductive amination. Such multiple binding causes aggregates to form, and results in low specific activity of radiolabel in the prepared conjugate mixture. Accordingly, coupling agents are needed that do not cause aggregate formation.